14:09:33 <markwash> #startmeeting glance
14:09:34 <openstack> Meeting started Thu Mar 27 14:09:33 2014 UTC and is due to finish in 60 minutes.  The chair is markwash. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:09:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:09:38 <openstack> The meeting name has been set to 'glance'
14:09:43 <markwash> hi glance folks
14:09:47 <markwash> sorry for the delay!
14:09:48 <rosmaita> hi
14:09:49 <ameade1> o/
14:09:50 <jokke_> hey
14:10:01 <arnaud__> o/
14:10:10 <nikhil__> o/
14:10:22 <hemanth_> o/
14:11:10 <markwash> so I've got  one update and then I think we should open up to some more topic suggestions
14:11:20 <markwash> #topic opening up juno development
14:11:53 <markwash> we are *so* close to opening up juno development, which will mean that feature work in master will no longer be frozen
14:12:01 <ameade1> wooo
14:12:09 <markwash> There are a few blockers I know of
14:12:17 <markwash> https://review.openstack.org/#/c/75898/
14:12:37 <markwash> https://review.openstack.org/#/c/75865/
14:13:45 <markwash> it looks like we might also need to address https://bugs.launchpad.net/glance/+bug/1298039
14:13:53 <markwash> but that should be relatively easy (?)
14:14:02 <markwash> (I'm just seeing it for the first time now)
14:14:21 <zhiyan> https://review.openstack.org/#/c/83007/
14:14:36 <markwash> then we must land https://review.openstack.org/#/c/83133/ to give the signal to ttx to cut the release and open up juno
14:14:49 <markwash> thanks zhiyan
14:15:01 <zhiyan> markwash: thank you
14:15:27 <zhiyan> markwash: and probably https://review.openstack.org/#/c/80674/
14:15:53 <markwash> once we open up juno, if there are any majorly complex changes, it might be worth waiting a little bit of time (one or two weeks) before we land them, just because it will be easier to backport fixes if we find any major bugs in the next week
14:15:56 <arnaud__> +1 for the last one
14:17:01 <markwash> okay looks like there is a bit more to be discussed around that last one
14:17:21 <markwash> for the moment, any more questions about opening up juno development?
14:18:13 <markwash> looks like a no, okay!
14:18:21 <markwash> #topic agenda spike
14:19:05 <markwash> I saw a note about tasks and taskflow on the agenda, do folks want to talk about that now, or defer to the later time and just report back on the discussion?
14:19:23 <nikhil__> not sure if harlowja_away is here yet
14:19:47 <markwash> okay cool, we can just wait. I'll try to at least lurk during that, if you guys have it in the #openstack-glance channel
14:20:00 <markwash> other agenda suggestions
14:20:32 <nikhil__> I have one
14:20:37 <jokke_> Can we discuss a bit about tags?
14:20:45 <nikhil__> may be ameade1 can pitch in
14:20:52 <nikhil__> it should not take long
14:20:57 <flwang> jokke_: any question about tags?
14:21:38 * markwash jots down
14:21:44 <jokke_> flwang: yeah few ... they kind of fell on my lap as we have a person doing some v2 testing
14:21:55 <nikhil__> it's about incorrect response for some of the calls. like HEAD on /images/details giving 500 <- courtesy ameade1
14:22:06 <markwash> any other items before we jump in?
14:22:40 <arnaud__> would like to reiterate on the nova bp in gerrit :)
14:23:02 <flwang> sorry for the break in, but it would be nice if we can land task client code in https://review.openstack.org/#/c/48850/
14:23:27 <ameade1> lol ok so what are we talking about?
14:23:38 <markwash> ameade1: atm we're just listing things we *want* to talk about
14:23:46 <zhiyan> markwash: i would like to know folks' toughs around auto glance.conf.sample generation
14:24:16 <zhiyan> s/toughs/thought
14:24:32 <ameade1> zhiyan: +1 i have some thoughts around that
14:24:42 <nikhil__> wow, seems like we should have a pre-meeting to jot all this down to the agenda etherpad
14:24:58 <zhiyan> ameade1: folks: https://review.openstack.org/#/c/83327/
14:25:27 <ttx> the psutil thing shall be fixed in requirements and then synced back. If you confirm it i can approve https://review.openstack.org/#/c/81373/
14:25:50 <ttx> markwash: there is a thread on it on openstack-dev fwiw
14:25:52 <markwash> okay, I added some notes to our agenda so we can roll forward
14:25:57 <markwash> ttx: thanks for the heads up
14:26:08 <markwash> #topic image tags (jokke_)
14:26:16 <markwash> jokke_: care to introduce the topic?
14:26:44 <jokke_> yeah
14:27:15 <jokke_> So we are testing Image API v2 and and our testing person came back to me telling that all his tag related tests failed
14:27:50 <jokke_> so I have been digging into it past few days and realized that the tags operations in glance/api/v2/images.py were not implemented
14:28:36 <jokke_> I've trying to get those done, but effectively it seems that all the tests touching tags are treating them as regular property so they fell apart
14:28:56 <markwash> ah I see
14:29:12 <markwash> jokke_: so PATCH was added after tags, and didn't work for them yet
14:29:21 <jokke_> markwash: yes
14:29:24 <markwash> we've been slowly moving the support for PATCH out to other attributes
14:29:36 <markwash> but were there failures other than PATCH?
14:30:13 <flwang> jokke_: good catch
14:30:30 <jokke_> The first was the set() passed to jsonutils.dumps()
14:30:39 <jokke_> I have WIP https://review.openstack.org/#/c/83046/
14:31:12 <jokke_> my local changes seems to work with glance client well at the moment, but breaks the old tests
14:31:53 <markwash> jokke_: would it make sense to separate the set() serialization fix from the patch functionality additions?
14:32:18 <markwash> it seems like fixing set() serialization is a bug fix rather than an additional feature, which makes it higher priority at this point
14:32:36 <jokke_> So one question I had was do we want to be able to change the tags at once with something like {'op': 'replace', 'path': ['tags'], 'value': ['ping', 'pong']}
14:32:52 <jokke_> markwash: I can split it out to get it landing asap if wanted
14:33:28 <markwash> jokke_: one other note on that front, I think it would be great if we *weren't* passing a set to the wsgi logic. Rather whatever serializer class that is returning a set right now should return a list I guess
14:33:58 <markwash> so a slight tweak to that fix may be in order, if others agree
14:34:35 <markwash> jokke_: I think we probably should support replacing /tags completely, if there are no particular difficulties you've seen?
14:34:54 <rosmaita> jokke_: what is the problem, again?
14:35:04 <rosmaita> PATCH for tags used to work
14:35:05 <jokke_> markwash: I agree, but bit like with the MultiDict I prefer having the sanitizer there even we would not hit on it even accidentally for next weeks/months instead of getting 500 out of the server
14:35:18 <ameade1> rosmaita: +1 i thought replace worked for tags
14:35:35 <rosmaita> replace definitely does
14:35:40 <jokke_> markwash: I believe it could be easier to implement than cange all the tests around there
14:35:44 <markwash> jokke_: fair point
14:37:11 <markwash> rosmaita: okay, so is there any bug currently with replace?
14:37:22 <rosmaita> no, don't think so
14:37:25 <markwash> okay phew
14:37:41 <jokke_> and please forgive me for any ignorance ... I'm new guy on the loop. Name is Erno Kuvaja and I started to work with Glance full time just 3 -4weeks back
14:37:52 <jokke_> rosmaita: I certainly doubt so
14:37:57 <ameade1> jokke_: welcome!
14:38:07 <markwash> all right I think that's settled then, fix the serializer and the wsgi serialization as bugs, add /tags/X support to PATCH in juno
14:38:12 <markwash> jokke_: welcome!
14:38:13 <arnaud__> welcome :)
14:38:31 <jokke_> rosmaita: As there is comment on the code that the tags part is not implemented yet
14:38:34 <jokke_> ;)
14:38:41 <rosmaita> so you can't "add" /tags because it's a protected element
14:38:47 <rosmaita> you cannot delete it, either
14:38:56 <rosmaita> but you can replace the /tags with []
14:39:04 <rosmaita> which deletes all the tags
14:39:31 <markwash> let's not get too bogged down now with what could just be bug reports
14:39:53 <markwash> nikhil__: I think you had a brief note about some 500 errors you found?
14:40:18 <nikhil__> well, ameade1 found HEAD on /images/details giving 500s
14:40:37 <ameade1> #link https://bugs.launchpad.net/glance/+bug/1297358
14:40:42 <nikhil__> I'd a suggestion around different calls for Glance API giving back Allow header
14:40:57 <nikhil__> like the above one
14:41:21 <nikhil__> it stems from the MP https://review.openstack.org/#/c/78050/
14:41:21 <ameade1> it would be cool to really iron out the api responses without adding too much complexity
14:42:09 <markwash> it sure would be great if routes mapper would handle that for us
14:42:10 <nikhil__> this is not a bug (on a bigger picture) however, a good usability experience
14:42:21 <ameade1> markwash: +1
14:42:24 <markwash> the 500 error is probably some bug somewhere in our wsgi stack
14:42:27 <nikhil__> yeah
14:43:06 <markwash> it looks like this bug report is in order, can you add a comment about the allow header, nikhil
14:43:14 <nikhil__> sure
14:43:18 <markwash> it looks like the Allow header is required for a 405 so we'd better make sure its there
14:43:33 <markwash> next!
14:43:44 <markwash> #topic gerrit blueprints (arnaud__)
14:44:32 <arnaud__> yeah: so did you guys got the time to look at what is happening for nova? what are your thoughts about it?
14:45:06 <arnaud__> context: nova is going to use gerrit+launchpad for blueprints: gerrit as a way to iterate through the design
14:45:12 <zhiyan> arnaud__: is that bp template really to use now?
14:45:18 <arnaud__> not yet
14:45:27 <arnaud__> several patches are out for review
14:45:33 <arnaud__> to modify it
14:45:34 <markwash> I skimmed throught that ML thread but didn't come out with as clear a picture as I wanted
14:45:38 <arnaud__> but I think it is not too far
14:45:54 <ameade1> would they be using that process in juno?
14:45:57 <markwash> (I think I just got lost in the comments)
14:46:00 <arnaud__> ameade1: yes
14:46:10 <ameade1> should we let them try it out for awhile and work out the kinks?
14:46:38 <jokke_> +1
14:46:46 <arnaud__> ameade1: sounds reasonable :)
14:46:53 <zhiyan> i was going to try to prepare one (for 33409 stuff), but seems other comments say stop
14:47:09 <markwash> ameade1: +1 from me
14:47:21 <arnaud__> ok! so seems we have an answer :)
14:47:24 <markwash> :-)
14:47:40 <markwash> #topic task code for the client (flwang)
14:48:11 <flwang> markwash: ameade has +2ed, so I need another bless on that
14:48:23 <flwang> and it would be nice if anybody can help verify that :)
14:48:26 <markwash> flwang: can you share the link?
14:49:14 <flwang> https://review.openstack.org/#/c/48850/
14:50:02 <ameade1> quick thought, is v2 the default version in glance yet?
14:50:07 <markwash> do folks feel like the task stuff in the api is in the state where exposing it in the client makes sense?
14:50:10 <ameade1> or 'current' version?
14:51:09 <rosmaita> YES!
14:51:13 <ameade1> markwash: i dont think it adds much more confusion than having it in the api
14:51:26 <rosmaita> ameade1: +1
14:52:14 <markwash> seems like a consensus of sorts. . if there are no objections then I'm fine with it
14:52:51 <markwash> sounds like there are none :-)
14:52:58 <markwash> next up!
14:53:08 <markwash> #topic config generation (zhiyan)
14:53:21 <markwash> zhiyan: can you give a little background and then I want to ask a question about it
14:54:19 <zhiyan> ok, time always tight. i listed the benefits in the commit message: https://review.openstack.org/#/c/83327/
14:55:35 <ameade1> jgriffith was working fairly late yesterday due to jenkins using a different version of the check sample config script which completely broke the gate for cinder
14:55:44 <ameade1> also, there is a ML thread about this too
14:56:24 <ameade1> i'm not to read up on either though
14:56:44 <markwash> zhiyan: so my question is, how do we deal with different configs for different executables ?
14:57:24 <markwash> I'm not very familiar with how the config generation works
14:57:36 <zhiyan> markwash: good point. from nova and cinder (probably other projects), seems deployer need to take care that by themself.
14:58:03 <ameade1> #link http://www.mail-archive.com/openstack-dev@lists.openstack.org/msg20428.html
14:58:23 <markwash> ameade1: thanks
14:58:26 <zhiyan> ameade1: thanks, i just can't find it out ..
14:58:36 <ameade1> #link https://review.openstack.org/#/c/83233/
14:59:29 <ameade1> i *think* what happened with cinder yesterday is that config options changed in keystoneclient and that changed what the sample config looked like so the sample in master was out of sync
14:59:51 <ameade1> i'd like to avoid issues such as this in glance so i just want to make sure we understand them
14:59:56 <markwash> gotcha
15:00:06 <zhiyan> and i think even we go "sample config files should be ignored " way, the change #83233 is fine since i added genconf testenv for tox
15:00:33 <zhiyan> ameade1: yes, keystoneclient 0.7.0 updated configs
15:00:48 <markwash> zhiyan: It seemed like nova has just one config file and you are supposed to sort out for yourself how to split it up among the nova services, is that right?
15:01:18 <zhiyan> markwash: time's up. we can talk this more in team irc?
15:01:56 <markwash> let's close out for now in any case
15:02:00 <markwash> thanks folks! sorry I was late
15:02:06 <markwash> at least I was in the right room this time! :-)
15:02:07 <jokke_> Thanks
15:02:09 <markwash> #endmeeting